From Feedback to Uptake: EFL Students' Perception and Use of ChatGPT Across Writing Stages

• In Event: AI in Higher Education: Balancing Support, Ethics, and Agency

Abstract

This study investigates how Korean EFL university students interact with ChatGPT during a structured academic writing task and the extent to which they integrate AI-generated feedback. Using a convergent mixed-methods design, the study analyzes interaction logs and interviews from 13 students. Quantitative results reveal varying levels of feedback integration across writing stages, with the highest in proofreading. Chi-square tests show significant associations between writing subprocesses and integration levels. Qualitative findings indicate students appreciated ChatGPT’s support for fluency and coherence, but raised concerns about superficial evidence and ethical ambiguity. Many adopted moderate integration to balance usefulness and authenticity. The study emphasizes the need for pedagogical models that promote critical, stage-specific, and ethical engagement with AI in EFL writing.

Authors

• Jiyoon Jung, Korea University

• Sewon Joo, Florida State University

• Insook Han, Korea University